Ion implantation is a process of depositing chemical species into a substrate by direct bombardment of the substrate with high-energy ions. In semiconductor fabrication, ion implanters are used primarily for doping processes that alter the type and level of conductivity of target materials. A precise doping profile in an integrated circuit (IC) substrate and its thin-film structure is often crucial for the IC performance. To achieve a desired doping profile, one or more ion species may be implanted in different doses and at different energy levels. A specification of the ion species, doses and energies is referred to as an ion implantation recipe.
FIG. 1 depicts a prior art ion implanter system 100. As is typical for most ion implanter systems, the system 100 is housed in a high-vacuum environment. The ion implanter system 100 may comprise an ion source 102 and a complex series of components through which an ion beam 10 passes. The series of components may include, for example, an extraction manipulator 104, a filter magnet 106, an acceleration or deceleration column 108, an analyzer magnet 110, a rotating mass slit 112, a scanner 114, and a corrector magnet 116. Much like a series of optical lenses that manipulate a light beam, the ion implanter components can filter and focus the ion beam 10 before steering it towards a target wafer 118. For illustration purposes, these components are hereinafter referred to as “beam-line elements.”
In semiconductor manufacturing, an ion implanter system often has to process many batches of wafers based on various recipes. For wafer batches processed with a common recipe, it is critical that the ion implanter system maintain a consistent ion beam output. Thus, at the beginning of each batch, it is necessary to tune the ion implanter system to reproduce an ion beam that is substantially the same as produced in the past batches with the same recipe.
However, due to a number of deficiencies, existing methods for tuning an ion implanter system often fail to achieve a consistent ion beam output in an efficient manner. For example, to repeat a previously successful ion beam setup, it is typical for existing tuning methods to rely on a single set of beam-line element settings recorded in the previously successful setup. Yet, many factors in the ion implanter system can change the ion beam condition even if the beam-line element settings are maintained the same. For instance, an ion source usually has a lifetime during which the ion generation gradually degrades. Therefore, even with identical beam-line element settings, the ion beam current can be quite different depending on the length of time the ion source has been in use. Since the previously recorded single set of beam-line element settings often cannot reproduce a desired ion beam condition, the ion implanter system has to be re-tuned for every wafer batch, which is quite time-consuming for reasons described below.
Existing tuning methods also tend to use ion beam current as the only criterion to optimize the ion beam. However, identical ion beam currents do not necessarily guarantee identical ion beam conditions. For example, identical ion beam currents can be produced with several different combinations of beam-line element settings. These different combinations often cause different ion beam dimensions, positions and angles. As a result, single-parameter approaches that rely solely on ion beam current can lead to inconsistent ion beam geometries. Further, due to different ion beam geometries, extra time must be spent, for each batch, on ion beam measurement, parallelism setup, uniformity setup, and implant dose control.
In existing tuning methods, the beam-line elements are usually adjusted, one at a time, to maximize the ion beam current. Since each beam-line element may have several settings, an ion implanter system with multiple beam-line elements can have numerous combinations of settings. The one-element-at-a-time approach is effectively a blind search in a vast data universe and therefore is very time-consuming. This approach also tends to ignore the correlations among the various beam-line elements. Since a small change in one beam-line element often results in a shift of an optimal point in another, the one-element-at-a-time approach can easily mistune the ion implanter system.
In view of the foregoing, it would be desirable to provide a solution for tuning an ion implanter system which overcomes the above-described inadequacies and shortcomings.